RETAIL

By Chris O'Neal -- John Picklo, Manager for High-Performance Computing at DaimlerChrysler, will kick off the ClusterWorld Conference with an opening keynote address Tuesday, June 24 at 9 a.m. Mr. Picklo is responsible for systems software and hardware for all of the engineering mainframes and supercomputers at the Chrysler Group. His background includes 25 years of experience working with information technology in various technical and consulting roles. Mr. Picklo's automotive background includes experience designing and managing systems to support computer-aided design at DaimlerChrysler, General Motors, Nissan, and Toyota. Supercomputing Online interviewed Mr. Picklo to learn more. 1. Please give our readers a brief overview of HPC at DaimlerChrysler. Please let me frame this conversation. I am speaking only for the Chrysler Group, rather than all of DaimlerChrysler. High Performance Computing (HPC) at the Chrysler Group seeks to enable our advanced engineers who simulate vehicle dynamics such as Impact Events, Fluid Dynamics, and Noise-Vibration-Harshness. HPC provides the computing infrastructure, in the form of hardware and software, to perform the simulations in a virtual world in lieu of building physical vehicle prototypes. Over the years, we have deployed various forms of supercomputers, including Vector based systems, Risc processors, and most recently, clusters of Intel based processors to provide simulation computing capacity. Specifically, in the past year, we have implemented both Pentium/Xeon and Itanium-2 based clusters to address 32-bit and 64-bit processing needs. These clusters have successfully provided a significant improvement in price-performance for simulation computing. 2. What are the major challenges? At a high level, we are faced with two major challenges - creating more desirable vehicles and lowering overall costs. HPC and cluster technology helps us address both challenges. The costs of clustered systems are considerably lower than their Risc-based predecessors. At the same time, performance has increased significantly. Performance translates directly into the ability to evaluate more design alternatives within the same design window. Being able to evaluate more designs allows us to deliver better vehicles while maintaining speed-to-market. 3. You’re the keynote speaker at the upcoming ClusterWorld Conference (June 23-26). Please give us a brief preview. The Chrysler Group has been running clusters in production since 1998. We have lived with a few different cluster configurations. At this time, we have three large clusters in production and may implement one or two more in the near future. My talk will attempt to share some of the knowledge that we have gained through our experiences. Specifically, I will be discussing implementation issues, cluster design alternatives, workload management, reliability, and our view of the near term future for clustered computing. Supercomputing Online wishes to thank John Picklo for his time and insights.